FINANCIAL IMPACT ANALYSIS METHODOLOGY
An overview of Climate Alpha’s data engineering, machine learning, and analysis method.
ABSTRACT
The increasingly unpredictable patterns and implications of climate change have made it more urgent than ever to understand the risks associated with various climate perils. From rising temperatures and sea levels to unprecedented wildfires, the threats posed by both acute and chronic climate events not only affect our natural and built environments but also the very fabric of our socio-economic structures. In response, Climate Alpha has developed industry-leading models for assessing physical risk and resilience – the capacity to recover quickly from these disruptions – and quantifying the impact these risks pose to real assets.
The objective of this white paper is to illustrate the methodology Climate Alpha employs to calculate the financial impact of climate change. We start by first calculating risk indicators for each climate peril as well as the community resilience as expressed through a range of socio-economic indicators. We then illustrate how these distinct, yet interconnected indicators come together to form a financial impact coefficient that provides a quantitative premium or discount investors can apply to an asset, enabling investors to make informed data-driven decisions, channel resources efficiently, and ultimately build a more resilient asset portfolio.
OVERVIEW
Climate Alpha reduces the complexity of interpreting and responding to climate volatility by generating meaningful data useful for location analysis and investment strategy. We begin by gathering raw, unstructured climate, socio-economic, and market data from an array of public and private sources. This data is meticulously engineered and refined into curated features that offer distilled insights into the impact of climate change.
Sets of related features are then consolidated into thematic indicators for calculation and analysis. These thematic indicators are bucketed into two main categories: Risk and Resilience. Risk Indicators assess the potential damage from physical climate hazards for any global coordinate based on historical data and future projections under multiple climate change scenarios (SSP2-4.5, SSP3-7.0, SSP5-8.5). Resilience Indicators assess a location’s readiness to absorb shocks as measured through a holistic range of attributes that are historically correlated to market performance.
These clusters of risk and resilience indicators are then fed into a pipeline that aggregates them into two quantitative coefficients: Physical Impact Coefficient and Resilience Adjusted Impact Coefficient. Physical Impact projects the net financial impact of physical climate risks on a given location at a given time in the future and under a given climate change scenario and is expressed as a percent change from a location’s projected market growth rate. Resilience-Adjusted Impact modifies the Physical Impact coefficient with the location’s resilience profile in order to account for its capacity to offset physical climate risks. (It is also represented as a percent change from that location’s projected growth rate.)
CLIMATE MODEL DOWNSCALING
The first step in our process is to understand the implications of physical climate risks at the most granular level. However, one of the primary limitations of global climate models is their wide resolution, typically ranging from 50km to 250km, making them inadequate for assessing localized impacts at the zip code, census tract or building level. To address this challenge, we employ machine learning (ML) to derive downscaled climate models that enable our platform to deliver up-to-date, high-resolution local climate impact simulations without hefty computational costs. (These methods also ensure our ability to seamlessly integrate the next iteration of global climate models.)
RISK & RESILIENCE INDICATORS
We begin by aggregating risk and resilience datasets (features) into thematic indicators. To understand the physical risk, approximately two dozen curated climate risk features from authoritative sources are engineered by Climate Alpha and aggregated into six thematic risk indicators: Heat, Drought, Inland Flooding, Coastal Flooding, Hurricane Wind, and Fire. (The methodologies of the aggregation of each risk indicator are detailed in their respective whitepapers). Because the underlying features have different scales, we standardize them using a Min-Max normalization for aggregation into a thematic indicator. The Total Physical Risk Indicator captures the aggregated distribution of risk across the six thematic indicators by summing each of the underlying indicators together. Next, we calculate the Z-score of each location to obtain its Relative Risk Profile for financial impact calculation.
To understand the resilience of a location, nearly 20 curated features from authoritative sources are engineering by Climate Alpha and aggregated into 6 thematic resilience indicators: Infrastructure, Energy Transition, Energy Reliability, Social Robustness, Economic Momentum, and Location Wellness (the methodologies of the aggregation of each resilience indicator are detailed in their respective whitepapers). Because the underlying resilience features have different scales, we standardize them using a Min-Max normalization for aggregation. The Total Resilience Indicator captures the aggregated distribution of resilience across six thematic indicators by summing each of the underlying indicators together. Next, we calculate the Z-score of each location to obtain its relative resilience profile for financial impact calculation.
FINANCIAL IMPACT CALCULATION
A location’s Impact Coefficients are determined by two factors: Relative Risk Profile and Climate Sensitivity. The calculation of a coefficient for a given location can be expressed as follows:
Relative Risk Profile measures the deviation of a location’s risk profile from the national mean profile. It is calculated by taking the Z-score (number of standard deviations by which an event is above or below the mean value) of a location against the risk profiles of all locations in the country. The more positive its Z-score, the more negative the physical climate impact. Conversely, the more negative its Z-score, the more positive the physical climate impact.
Climate Sensitivity measures the estimated accumulative economic damage as a percentage of GDP from future climate hazards. Based on existing scientific studies on the damage of climate change on local GDP1, each type of climate hazards costs roughly 1.2% of GDP per +1°C in average temperature.
We apply this value to calculate the climate sensitivity of a location in the year 2050, linearly interpolating downward to 0% in 2020. In the year 2050, if a location is exposed to a certain climate risk (with a score larger than 0), its sensitivity is increased by 1.2%, reflecting a -1.2% impact for each risk category. Thus, if a location is exposed to all 6 climate perils, its climate sensitivity is 6 * -1.2% = -7.2%. Similarly, a location exposed to 5 risks will have a climate sensitivity is 5 * -1.2% = -6%.
Putting the two factors together, if the Relative Risk Profile of a location has a Z-score of 2 with an estimated Climate Sensitivity of -6%, the Impact Coefficient for this location in the year 2050 will be (-6) * 2 = -12%. Thus, if the estimated value of a property is forecasted to be $10M in 2050, an impact of –12% yields a total reduction of $10M * 12% = $1.2M in asset value, leading to a risk-adjusted value of $8.8M.
PHYSICAL IMPACT COEFFICIENT
The figure below illustrates the process by which we form a location’s Relative Physical Risk Profile, which measures the deviation of a location’s physical risk from the national mean and is used to calculate the Physical Impact Coefficient.
The final Physical Impact Coefficient is then determined by multiplying a locations Relative Physical Risk Profile from Figure 3 by its Climate Sensitivity:
For example, if the Relative Physical Risk Profile of a location has a Z-score of 2.5 with an estimated Climate Sensitivity of 2.4% loss (-2.4%) in the year 2050, the Physical Impact Coefficient for this location in the year 2050 will be (-2.4) * 2.5 = -6%. If the estimated value of a property is forecasted to be $10M in 2050, a Physical Impact Coefficient of -6% yields a risk-adjusted value of $9.4M.
RESILIENCE-ADJUSTED IMPACT COEFFICIENT
Figures 9 and 10 below illustrate the process by which we form a location’s Relative Resilience-Adjusted Risk Profile, which modifies the Relative Risk Impact Profile and is used to calculate the Resilience-Adjusted Impact Coefficient. This paints a more holistic picture of how climate change impacts assets in locations by considering the ability of a location to offset physical risks.
For example, if the Relative Physical Risk Profile of a location has a Z-score of 2.5 with an estimated Climate Sensitivity of 2.4% loss (-2.4%) in the year 2050, the Physical Impact Coefficient for this location in the year 2050 will be (-2.4) * 2.5 = -6%. If the estimated value of a property is forecasted to be $10M in 2050, a Physical Impact Coefficient of -6% yields a risk-adjusted value of $9.4M.
For example, if the same location boasts robust resilience traits to offset physical climate risks, its Relative Resilience-Adjusted Risk Profile could now have a Z-score of 0.5 with the same Climate Sensitivity of 2.4% loss (-2.4%) in the year 2050. The new, Resilience-Adjusted Impact Coefficient for this location in the year 2050 will be (-2.4) * 0.5 = -1.2%, resulting in a projected 2050 valuation of $9.88M. The cumulative damage of $0.12M estimate is much lower than its Physical Impact Coefficient.